micro climates and moisture induced damage to paintings

Micro climates and moisture induced damage to
paintings
Marion F. Mecklenburg
Abstract
Because of cold climates and minimal insulation in the
walls of many buildings, the inside surfaces of exterior
walls reach the dew point. Very high humidity levels
and even condensation can occur on these surfaces. On
hot summer days these same walls can reach elevated
temperatures where the relative humidity near the
interior surface can drop to as low as 30%. The zone
close to such walls represents micro-climates that
can differ widely from the intended “controlled”
climates of museums and galleries. As a consequence,
considerable damage to works of art can occur. This
paper discusses some of the mechanisms that cause the
damage to canvas paintings.
Micro climates near the interior
surfaces of exterior walls
The most frequently encountered damage to
paintings, both on canvas and on wood, results
from exposure to high moisture levels. In both new
and old historic buildings, moisture can condense
on the inside of exterior walls from a variety of
reasons. One of those reasons is the excessive
humidification of the interior spaces of the building
in the wintertime.
Figure 1 shows condensing water leaking from
the exterior walls of the Smithsonian Institution’s
Figure 1. Condensed water leaking from the joints in the wall of
the Smithsonian Institution’s Hirshhorn Museum and Sculpture
Garden in the wintertime. (Photo by Gary Johannsen and Kevin
Guifreda)
Figure 2. A computer generated temperature profile of an
existing wall at the Hirshhorn Museum and Sculpture Garden.
The inside surface is to the right. [1]
Hirshhorn Museum and Sculpture Garden. This
water is a result of condensing humidified air
within the walls on cold winter days. The Hirshhorn
Museum and Sculpture Garden opened to the public
on Oct. 4, 1974, so it is a relatively new building.
Because of the architectural design of the building,
the structural walls are massive. None the less, the
walls get cold and the condensation occurs.
Figure 2 shows the temperature profile of an existing
wall at the Hirshhorn Museum. The reinforced
concrete structural wall is 892mm thick and where
there are two air spaces, there is no insulation. On
very cold winter days it can reach -10C outdoors
where the interior environment of the building is 21C
and 55% RH. While the temperature profile in Fig.
2 was generated by computer, actual measurements
of select spaces verified that the outside, inside and
inner wall temperatures are fairly accurate.
In the United States, many older buildings are
constructed of brick masonry, no insulation,
minimal air spaces, and highly decorative plaster
interiors. These buildings are often converted for
use as galleries and historic house museums. Some
have been retrofitted with sophisticated heating,
ventilation and air conditioning (HVAC) systems. In
the winter time, the exterior walls of older buildings
can get quite cold and the interior surfaces can
reach the dew point because of elevated indoor RH
levels.
Museum Microclimates, T. Padfield & K. Borchersen (eds.) National Museum of Denmark 2007
ISBN 978-87-7602-080-4
19
Figure 4. The construction of a traditional canvas supported
painting. This assembly includes the “support” canvas, a
glue size layer, an oil ground and the oil design layers. This
particular section was from a 19th century Italian painting. The
green bar at the top of the picture is 1mm . (The cross-section
and photograph courtesy of Melvin J. Wachowiak)
Figure 3. Condensed moisture running down the wall from
behind one of several paintings at the Renwick Gallery of
the Smithsonian Institution in Washington, D.C. (Photograph
courtesy of Ehrenkrantz, Eckstut & Kuhn, Architects)
Figure 3 shows condensed water running down the
interior face of an exterior wall from behind a painting
at the Renwick Gallery of the Smithsonian Institution.
The walls are 660 mm thick brick masonry walls
having a 25 mm air space and a 25mm plaster finish.
There is no insulation. The painting acted as insulation
along the lower edge where there was contact with the
wall. Moisture condensed at this lower edge. In this
case the outdoor temperature was around -10°C and
the temperature and relative humidity of the interior
spaces was 21°C and 50%, with a dew point of 10°C .
In the summer time, the exterior walls get hot and
the space behind the painting can be warmer than
the interior of the exhibition gallery. If the outside
wall surface temperature is 35°C (or hotter if in
direct sunlight) there is a potential for the inside of
the wall to reach a temperature of 28°C or more.
In such cases the relative humidity behind the
painting drops and can get as low as 33% or less.
The microclimate behind paintings hanging on the
inside of an exterior wall is entirely different from
the “controlled climate” in the center of the gallery
space. Behind paintings hanging on exterior walls it
is entirely possible to have annual RH fluctuations
from 30% to 100% while the center of the interior
space is maintaining a constant 50% RH.
layers together. A cross-section of a traditional
19th century Italian canvas supported oil painting
is shown in Fig. 4. This assembly includes the
“support” canvas, a glue size layer, an oil ground
and the oil design layers. The glue size layer is
almost too thin to see.
Figure 5 shows a detail of the same 19th century
Italian painting but looking from the front. The glue
size layer is an extremely thin film bridging the gaps
in the weave of the canvas. Even though very thin,
this layer is still very responsive to changes in RH.
It is commonly assumed that the canvas is the support
of an old master painting but in fact the glue size
provides support over most of the RH range. [2] This
can be illustrated by looking at the individual layers
of the painting when they are restrained and subjected
to changes in relative humidity. While exploring
each layer’s response to environmental changes it
is possible to determine the actual mechanisms that
cause damage at different levels of relative humidity.
It is possible to measure the magnitude of forces in
restrained linen samples as the RH changes. Note
that the force per unit width acting on individual
materials is used in this paper, since it is not
The effects of cycling rh on canvas
paintings
Canvas paintings represent some of the most
complex structures in the cultural world. This is
because of the widely varied materials used and their
independent response to the environment. This can
be illustrated by looking at each layer of a typical
painting individually and then superimposing the
20
Figure 5. A detail of the same 19th century Italian painting
shown in Fig. 4 but looking from the front. (Photograph courtesy
of Melvin J. Wachowiak)
practical to calculate the stresses in textiles as force
per unit area. Also, each of the materials examined
will be of the thickness that might be encountered in
typical easel paintings. Using this strategy, it is also
possible to include the effects of the thicknesses of
each of the different layers.
In building the composite painting up from the
support canvas, it is useful to start with the canvas.
The sample of linen tested was from an Ulster #8800
canvas. It is a medium weight canvas such as would
be found on many easel paintings. Both the warp
and weft directions were tested. An initial force was
applied to the specimens at mid RH and the relative
humidity was incrementally changed while the
new level of force per width was recorded. As the
relative humidity changed, so did the force acting
on the canvas. This was continued for several cycles
over a large range of relative humidity.
Figure 6 shows that between 10% RH and 60% RH
there is relatively little change in the force on either the
warp of fill directions of the textile. From 70% RH on
there is a gradual increase in stress and above 80% RH
the force increases dramatically. When damp or wet,
loose textiles shrink dramatically and when restrained
the shrinkage shows up as significant forces in the
textile. This is the first indication that dramatic events
take place in canvas paintings when the humidity gets
very high. This behaviour was replicated using a wide
variety of different textiles by Gerry Hedley at the
Canadian Conservation Institute. [3]
Of all of the materials used in canvas paintings, hide
glue is the strongest and nearly the stiffest. It is also
the one material that develops the most force when
restrained and desiccated. It is because this material
is both stiff (and strong) and has a high dimensional
response to humidity that it develops so much force.
Figure 7 shows the force per width (stress) developed
Figure 6. The tensile forces per width, measured in individual
restrained samples of the #8800 linen in the warp and fill
directions with changing relative humidity. The greatest forces
develop when the relative humidity is above 75%.
Figure 7. The force per width of restrained samples of hide glue
when desiccated from 85% to 15% RH. The stress developed in
the hide glue at the maximum force per width of these samples
was 27 MPa.
in a very thin film of glue, 0.012mm, when it is
restrained and desiccated from 85% to 15% RH. From
80% RH and above, the hide glue has no strength and
therefore no ability to maintain the bond between the
canvas and ground layers. The thickness of the film
is about the same as that found as a size coating on
paintings. (See Figures 4 and 5)
In general, the force per width developed in restrained
and desiccated oil paint is considerably less than
the force per width developed in other materials
found in paintings. One of the reasons is that with
the exception of some of the paints made with the
earth colours, the dimensional response to humidity
changes is low. While the earth colors tend to have
a higher dimensional response, they have relatively
low stiffness. Figure 8 shows two paint samples
restrained and desiccated from around 75% to 5%
RH. Even with this large change in relative humidity,
the forces developed are low. So the likelihood that
large excursions to low humidity can damage the
oil paint layer is low. It takes a combination of
Figure 8. The force per width of restrained samples of lead white
and Naples yellow oil paints. The stress of the white lead paint
at the maximum force per width of this sample was only 0.65
MPa. The force per width of the paints is considerably lower
than the hide glue and a bit higher than the #8800 linen shown
in figure 6. The thicknesses indicated for the paint samples is
typical of those found in paintings.
21
Figure 9. The force per width of restrained samples of linen,
hide glue, and lead white and Naples yellow oil paints. The
thickness of these films are as shown in their individual figures
and would be typical of an oil painting painting. Using this
figure it is possible to compare the responses of the individual
layers to that of an actual canvas painting and to determine the
different forces occurring at different levels of RH.
materials and their individual responses to changes
in humidity to cause deterioration. This can be
demonstrated by superimposing all of the layers of
a painting together and comparing the results with
an actual painting.
Superposition of the different layers of
a painting
The information from Figs. 6, 7, and 8 is plotted
on the same graph in Fig. 9. This makes it possible
to compare the responses of the individual layers
of a canvas painting to RH and to identify the RH
levels that hold the most potential for damage. For
example the fabric is developing high forces only
at high RH levels and staying relatively constant
at humidity levels below 80%. The hide glue is
developing high forces at very low RH levels but
loses all strength at levels above 80%RH. The paint
films are developing relatively low forces and that
is only at very low levels of RH.
The restrained testing of samples from
actual paintings
Figure 10 shows the force per width developed in
restrained samples of a 1906 painting by Duncan
Smith. This painting was constructed with a
medium weight, machine woven linen fabric, a hide
glue size, a lead white ground and a design layer
of raw and burnt umber. It is important to note that
there are two areas of high force development,
one at the very low levels of RH and the other at
the very high levels of RH. This is consistent with
the force development of hide glue and canvas as
22
Figure 10. The forces per width of restrained samples of an
unknown American portrait painted by Duncan Smith (1906),
in both warp and weft directions. These painting samples were
constructed with a medium weight machine woven fabric, a
hide glue size, a lead white ground and a design layer of raw
and burnt umber.
shown in Fig. 9. In the actual painting the fabric is
developing high forces at high RH and the glue size
is developing high forces at low RH.
Above 80% RH the hide glue is no longer acting
as the secure bond between the ground and linen
canvas. From 80% RH and above the paint layer is
clearly at risk of delaminating from the canvas. At
this same RH the paints films are the most flexible
but are also in their weakest state. Above 80% RH,
the fabric will shrink if loose and certainly could
cause delamination of the design layers attached
to it. One further comment here is that from 10%
RH to 75% RH, the force level in the glue layer is
much higher than in the other layers, including the
linen canvas. In this range it is the hide glue that is
supporting the painting, not the canvas.
Not all linens show the same behaviour. Some
linens are woven such that the weft direction yarns
are quite straight and have little crimp. It is the
crimp in a yarn that causes high humidity shrinkage
when loose and high forces when restrained. Also
low quality linens can be used for commercially
prepared artists’ canvases. In order to get a stiffer
feel for the linen, heavier layers of glue size are
applied before the oil ground is applied. This results
in even higher forces in the low humidity ranges.
Effects of cycling canvas paintings over
large ranges of relative humidity
If a canvas painting is exposed to large cyclic changes
in relative humidity, a pattern of corner cracks can
occur. This can be demonstrated by constructing a
“mock” painting of canvas, a size layer of hide glue
and a “design layer” composed of a hard gesso film
having the mechanical properties of an old brittle
oil paint film. The dimensions of the painting were
505mm x 762mm. The gesso layer was a hide glue
and calcium carbonate mixture. [4]
Figure 11 illustrates the results of such an experiment.
This mock painting was cycled from 90% RH to 35%
RH and then back to 90% RH. Each half cycle (from
high to low RH or low to high RH) required just less
than 24 hours for full equilibration. Periodically, the
test painting was examined to see what cracking
might have occurred. It was observed that with
one small exception, all of the cracks occurred at
the corners of the painting. At the ends of selected
cycles (#4, #7, and #9), the ends of the cracks were
marked.
Figure 12. The combination of crack patterns from stretcher
expansion and RH cycling over a large range.
After nine full cycles the crack extension ceased,
as was confirmed by additional cycles. The painting
was then subjected to several more severe cycles
from 95% RH to 20%RH and back. There was no
additional cracking or crack extension. What is of
interest is that the first 4 cycles caused the most
damage and subsequent cycles only produced
smaller increments of crack extension until it ceased
altogether. More severe RH cycles did not add to the
damage. The cracks that did occur began to act as
expansion joints and now the painting can experience
large RH cycling without further damage.
the glue layer acting on the paint layer that causes
damage. The cracks shown in the corner of the test
painting (Fig 11) reflect the effects of the hide glue
(and to a small extent the paint itself) pulling from
the central regions and away from the corners of the
painting. This distorts the design layers sufficiently
to cause cracking in the paint layer at the corners.
From the discussion above, hide glue loses strength
at high humidity levels but develops very high
stresses when desiccated. It was also shown that,
acting alone, paint layers won’t generally develop
high stresses and damage themselves when
restrained and desiccated. It is the desiccation of
Condensed moisture damage to canvas
Figure 11. The results of cycling an experimental “mock”
painting to cycles of large changes in relative humidity. The cycle
range was from 90% RH to 35% RH. The crack limits at 4, 7 and
9 cycles are marked. Additional cycling of a greater RH range
beyond the initial nine cycles did no further damage, because
the cracks relieved the stresses caused by further RH cycles. This
model painting was constructed with a stretched canvas, a hide
glue size and a stiff gesso acting as a design layer.
In an actual painting, it is not unusual to see both the
cracking from stretcher expansion and environmental
cycling in large ranges of relative humidity combined.
This is illustrated in Fig. 12. [5]
paintings
One of the most frequently encountered types of
damage to paintings, both on canvas and on wood is
a result of exposure to high moisture levels. In old
historic buildings, the moisture can condense on the
inside of exterior walls from a variety of reasons.
One reason has already been explained. Another
reason that condensation can occur is that in old stone
buildings, the masonry walls are cooled during the
wintertime. These massive stone walls, due to their
high thermal mass, are slow to warm up with the
changing seasons and in the spring time warm moist
air enters the building along with visitors through
open doors. This results in extensive condensation
on not only the walls but on paintings hanging on
those walls. This occurs on many of the monuments
such as the Lincoln and Jefferson Memorials in
Washington, D.C. in the United States.
Water condensing on paintings often tends to run
to the bottom of the painting and typically causes
damage as shown in Fig. 13. In this case there
23
Figure 13. A detail of a 19th century Italian painting. It is clear
that total separation of the paint and ground layers from the
canvas has occurred. The amount of moisture was sufficient to
cause cracking of the design layers and failure of the bond at
the glue layer. The canvas shrank, and the paint cleaved from
the canvas. (Photo by Matteo Rossi Doria)
was sufficient water on the canvas that it totally
disrupted the adhesive bond of the animal glue size
layer. Hence the canvas shrank, glue size lost all of
it adhesive strength and the paint and ground layers
completely detached from the canvas. Now there
is insufficient room to fit the broken pieces of the
paint back into proper alignment.
Selectivity in RH Damage
The chemistry of oil paint is very complex and,
understandably, properties of dry films vary with
the pigment. For example, paints made with basic
lead carbonate dry to tough durable films while
those made with the earth colors form weak films.
[6 7] The mechanical properties of several different
paints are illustrated in Fig 14. One would expect
that the white lead paint, because of its strength and
low dimensional response to moisture would survive
large swings in relative humidity. On the other hand
one might suspect that weak and dimensionally
responsive paints like umber, ocher, and Sienna
would most likely suffer considerable damage in the
same harsh environment. [8]
At high moisture levels the earth colours have
very little strength and the hide glue size has none.
Therefore neither has the ability to resist damage.
Failure is going to be preferentially in the earth
colours. Figure 15 illustrates selective damage to a 19th
century Italian painting. Here the white lead paints are
relatively intact while the earth colours are seriously
damaged. Clearly avoiding high humidity levels is
of primary importance. Moisture induced damage to
paintings is selective in that the weaker paints will
fail and the durable ones will maintain some stability.
24
Figure 14. The results of stress-strain tests conducted on paints
made with different pigments. The various pigments have a
dramatic effect on the mechanical properties of oil paints.
This is in contrast to damage due to excessively low
temperatures which has the same adverse effects on all
paints. [4, 5]
Conclusions
While it is possible to show that low relative humidity
excursions can cause damage, the most serious RH
related damage to paintings is generally a result of
very high moisture levels. While the source of the
moisture can be leaking roofs or damp basements,
a common cause of damage is high humidity and
condensation on the inside surface of cold exterior
walls. Lowering the indoor relative humidity and
keeping an air space between the works of art and the
wall can go a long way in minimizing damage. [9]
Author
Marion F. Mecklenburg, Ph.D.
Smithsonian Museum Conservation Institute
mecklenburgm@si.edu
Figure 15. A detail of a painting containing both white lead
paints (blue arrows) and paints made with the earth colours,
ochre and sienna (red arrows). This painting was damaged by
wet walls. (Photo by Matteo Rossi Doria).
References
1This plot was constructed from data taken from
“Basis of Design Report, Hirshhorn Museum
& Sculpture Garden Repair Exterior Structural
Leaks,” OFEO Project No. 0521104, Prepared
for the Smithsonian Institution and by HNTB
Architecture, (2006), p 81.
2Mecklenburg, M.F., “Some Aspects of the
Mechanical Behavior of Fabric Supported
Paintings,” Report to the Smithsonian
Institution, Research supported under the
National Museum Act, (1982)
3Hedley, G. “Relative Humidity and the Stress/
Strain Response of Canvas Paintings: Uniaxial
Measurements of Naturally Aged Specimens,”
Studies in Conservation, 33 (1988) 133-148.
4Mecklenburg, M.F., McCormick-Goodhart,
M., and Tumosa, C.S., “Investigation into the
Deterioration of Paintings and Photographs
Using Computerized Modeling of Stress
Development,” JAIC 33(1994) 153-70.
5Mecklenburg, M.F. and Tumosa, C.S.,
“Mechanical Behavior of Paintings Subjected
to Changes in Temperature and Relative
Humidity,” Art in Transit, Studies in the
Transport of Paintings, M.F. Mecklenburg, Ed.
National Gallery of Art, Washington, D.C.,
(1991), 173-216.
6Mecklenburg, M. F., C. S. Tumosa and D.
Erhardt, “The Changing Mechanical Properties
of Aging Oil Paints”, in Materials Issues in Art
and Archaeology, VII, Volume 852, Materials
Research Society (2005), 13-24.
7Tumosa, C. S., Erhardt, D., M. F. Mecklenburg
and Xingfang Su, “Linseed Oil Paint as
Ionomer: Synthesis and Characterization”, in
Materials Issues in Art and Archaeology, VII,
Vol. 852, Materials Research Society (2005),
25-31.
8Mecklenburg, M.F, “The Structure of
Canvas Supported Paintings,” Preprints of
the International Conference on Painting
Conservation, Canvases: Behavior,
Deterioration and Treatment,” Valencia, Spain,
March 2005, 119-155.
9 Mecklenburg, M. F., C. S. Tumosa and
A. Pride, “Preserving Legacy Buildings”,
ASHRAE Journal 46(6):S, 2004, 18-S23.
This work is licensed under a Creative
Commons Attribution - Noncommercial - No
Derivative Works 3.0 Licence.
25